Understanding water temperature variation in regulated rivers and reservoirs becomes increasingly important as the environment and ecosystem are approaching their thermal limits. In this paper, a multi-model approach is used to quantitatively access the spatio-temporal change in thermal structures of the Danjiangkou reservoir and its downstream river. The area is subject to thermal and hydrological alterations due to three large water diversion projects and related auxiliary projects, including a project to heighten the Danjiangkou dam and two small downstream reservoirs. It is found that the Danjiangkou dam heightening project alters water temperature seasonally, increasing it in winter and decreasing it in summer; while the three large water diversion projects and the two small downstream reservoirs mitigate the effect. Water temperature change in the downstream river is also studied from the aspects of release temperature and release discharge of the Danjiangkou reservoir. The former mainly changes the water temperature near the dam, while the latter affects the recovery rate and the recuperation distance. Ecological impact of the water temperature change is discussed based on the spawning of fish, indicating that the spawning periods may lag behind and the optimal spawning locations may move downstream.

INTRODUCTION

Water temperature is one of the most significant physical variables of water bodies affecting most water quality characteristics (Cluis 1972; Albek & Albek 2009; Hadzima-Nyarko et al. 2014). In addition, it plays an important role in the biological cycles of many aquatic organisms and changes in species composition (Watanabe 1998; Politano et al. 2008; Rayne et al. 2008; Hilderbrand et al. 2014). Generally, humans have changed the thermal structures in reservoirs and rivers by undertaking water conservancy projects, which may have an adverse influence on the environment and ecosystem (Smith 1975; Preece & Jones 2002). Hence, much attention has been paid to the influence of these projects on the water temperature distribution in reservoirs and the thermal regime in rivers (Webb & Walling 1993; Preece & Jones 2002). Previous studies have mainly focused on the effect of a single project such as a reservoir or a water diversion project (Meier et al. 2003; Khangaonkar & Yang 2008; Wang 2013). In fact, many reservoirs and river reaches are subject to multi-source thermal disturbances caused by several projects (Prats et al. 2010), but the effects of different sources of thermal disturbances have rarely been reported in the literature (Prats et al. 2012). Therefore, it is necessary to analyze the water temperature change caused by a series of projects in order to better understand their effects on the environment and guide project operation and water resources management in the future.

The causes of water temperature change due to water conservancy projects need to be researched in order to accurately evaluate and predict their effects on thermal regimes of water bodies. Many studies have discussed the water temperature variation of the downstream river caused by relevant projects with a reservoir from the aspects of either the release temperature or the release discharge. If thermal stratification occurs, water from deep-release reservoirs cools the river in summer and warms it in winter (Null et al. 2013). In addition, different operations of reservoirs can alter water temperature regimes because they change the release discharge and then affect thermal inertia in the river (Carron & Rajaram 2001; Meier et al. 2003; Khangaonkar & Yang 2008). However, these studies focused only on the direct effect due to these projects and ignored the indirect effect due to the thermal structure variation of the reservoirs. For example, when the influence on downstream river temperature by water diversion projects from the reservoirs is analyzed, it is often the effect of reduction in discharge (direct effect) that is considered. In fact, reducing discharge may change the thermal structure in the reservoir, affecting the release temperature and impacting the thermal regimes of the receiving rivers and streams (indirect effect). Therefore, both the water temperature structure in the reservoir and its downstream rivers should be studied to obtain a comprehensive and reasonable evaluation of the water temperature change in the river.

To quantify water temperature alterations caused by water conservancy projects, previous studies often used the concept of recuperation distance, which represents the distance needed to recover to natural state (Herb & Stefan 2011; Prats et al. 2012). In fact, recuperation distance is usually obtained based on some assumptions and simplifications, and cannot represent the real situation. In the study, recuperation distance is calculated based on the assumption that when the influence of release temperature from a reservoir on the stream temperature downstream is less than or equal to 5%, the river reaches its natural state. The advantage of the approach is that the calculation is simple as the stream temperature does not need to be obtained first.

Simulation and forecasting water temperature would substantially enhance analyzing the ecological impact of water conservancy projects. Previous studies have suggested that water temperature change will influence fish habitats in the river (Preece & Jones 2002; Bartholow et al. 2004) and agricultural activities in surrounding regions due to cool water release from deep reservoirs (Yang et al. 2012). Some indexes are used in these studies to assess whether the change is biologically critical. For example, Bartholow et al. (2004) used indexes including temperature degree-days, annual exposure, and macro-habitat suitability as relative guidelines to reveal biologically relevant differences between scenarios. As well, spawning opportunities index, which is the ratio of the time of water temperature above the minimum water temperature requirement for spawning against the whole spawning period, was applied to qualitatively evaluate Keepit Dam's ecological influence (Preece & Jones 2002). In this study, two indexes, including spawning opportunities index and the time to reach the minimum threshold for spawning are used to assess the possible ecological impact as they can judge spatial and temporal effects of the water temperature variation, respectively.

Danjiangkou reservoir and its downstream river is a typical system subject to multi-source thermal disturbances, including three large water diversion projects and related auxiliary projects. In this study, a series of scenario simulations involving different project operations are carried out based on a multi-model approach. The Environmental Fluid Dynamics Code (EFDC) model, which includes three-dimensional (3-D) hydrodynamic and water quality modules (Hamrick 1992; Jeong et al. 2010), is used to simulate reservoir temperatures. In addition, an analytical solution model based on the equilibrium temperature concept (ETM) model is used to estimate stream temperatures. ETM is a simple approach to estimate monthly average water temperature downstream of the dam based on monthly average release temperature, air temperature, and flow volume. More details about the ETM model are given in the section ‘Modeling approach’. The main objectives of this study are: (1) to quantitatively evaluate the thermal temporal-spatial change of the Danjiangkou dam and its downstream river; (2) to analyze the causes of the thermal regime change in the downstream river; and (3) to predict the ecological impact based on fish spawning due to these projects.

DATA AND METHODS

Study area

The study area includes Danjiangkou reservoir and its downstream river (Figure 1). The Danjiangkou reservoir (32°36′–33°48′N; 110°59′–111°49′E) is located on the borders of the Hubei province and Henan province, China. Its drainage area of approximately 95,000 km2 includes the upper Han River and the Dan River with backwater lengths of 177 km and 80 km, respectively (Zhang et al. 2009). During flooding periods (May to September), its outflow includes two parts, i.e., the water used for hydropower and surplus water from the overflow weir. Hence, the release temperature is affected by the temperature of both parts from May to September. The bottom elevation and the height of hydropower intake are 115.0 m and 7.5 m, respectively. The bottom elevation of the overflow weir is 138 m. The study stream reach downstream from Danjiangkou reservoir involves the middle and lower reaches of Han River, totaling 650 km. The study area has a north sub-tropical monsoon climate with rapid and severe climate transitions. Annual mean temperature is 15–16 °C. Precipitation is subject to large inter-annual variability, with 800–1,000 mm yr−1, of which 80% is concentrated between May and September (Li et al. 2008).
Figure 1

Map of the study area.

Figure 1

Map of the study area.

The Danjiangkou dam was heightened in 2010, from 157.0 to 170.0 m, as a precursor to the large water transfer projects that followed. Currently, there are three long distance water transfer projects being developed in the study area. The first is the South-to-North Water Transfer Project, which was completed at the end of 2014. 9.5 × 109 m3 of water are extracted every year from the Danjiangkou reservoir and delivered to Beijing, Tianjing, and other cities along the water transfer route. The second is the Ebei Water Transfer Project, with construction about to start in 2015. This will transfer 1.398 × 109 m3 of water per year from the reservoir to the Ebei area. The third is Han River-to-Wei River Water Transfer Project, which is scheduled for completion in 2015. It will deliver 1 × 109 m3 of water yearly from the Han River upstream of the Danjiangkou reservoir to the Wei River. To control the effect of these water diversion projects on the hydrological regime variability downstream of the Danjiangkou reservoir, there are two small reservoirs, finished in 2000 and 2010, respectively, known as Wangfuzhou and Cuijiaying. The two reservoirs are located 30 km and 134 km downstream from the Danjiangkou reservoir, respectively. Table 1 shows the physical characteristics of the two small reservoirs, alongside those of the Danjiangkou reservoir before and after heightening.

Table 1

Physical characteristics of the three reservoirs in the study area

Reservoir Normal water level, m Utilizable capacity, 106 m3 Hydropower capacity, MW 
Danjiangkou (heightened) 157 (170) 17,450 (29,050) 900 
Wangfuzhou 86.23 150 109 
Cuijiaying 62.73 245 96 
Reservoir Normal water level, m Utilizable capacity, 106 m3 Hydropower capacity, MW 
Danjiangkou (heightened) 157 (170) 17,450 (29,050) 900 
Wangfuzhou 86.23 150 109 
Cuijiaying 62.73 245 96 

Four species of Chinese carp, including Mylopharyngodon piceus, Ctenopharyngodon idellus, Hypophthalmichthys molitrix, and Aristichthys nobilis, are widely distributed in the middle reaches of the Han River (Danjiangkou reservoir to Zhongxiang City). There are five spawning sites accounting for 36.22% of the length of this reach (Li et al. 2006). The minimum temperature threshold for spawning in these species is 18 °C and the breeding period runs from late May to early August (Shi & Huang 2009). As water temperature variation due to these conservancy projects has affected fish habitats in the study reach, these fish can be used as a marker of ecological impact in the study.

Data source

For the study, topographical, reservoir operation, hydrological and meteorological data have been collected. The first three kinds of data were directly provided by the Yangtze River Water Resources Commission, and the last by the National Climate Centre (NCC) of China Meteorological Administration (CMA) and downloaded from the website: http://cdc.nmic.cn/.

Water temperature was monitored in the Danjiangkou reservoir during the period from 1969 to 1980. A monitoring section was set 200 m upstream from the dam, which consisted of six or seven vertical profiles, each containing nine to 16 measuring points. The distance between adjacent profiles was 110 m. Vertical profiles consisted of observation at several fixed water depths, including surface, 0.1, 0.5, 1.0, 2.0, 3.0, 5.0, 10.0, 15.0, 20.0, 25.0, 30.0, 35.0, 40.0, 50.0, 60.0 m, and bottom. The monitoring section recorded the water temperature variation of the whole cross section from the surface to the bottom and from the left bank to the right bank. Figure 2(a) shows the spatial distribution of the water temperature at the cross section on 15 August 1970. The section was observed once monthly at 8:00 a.m. on the 15th of the month. Another vertical profile, 50.0 m upstream from the dam, was also arranged for supplemental measurements and observed three to six times a month. This profile was close to the hydropower intake and hence could record its temperature variations. Figure 2(b) showed the water temperature variation at the hydropower intake during the period from 04/1973 to 03/1974. Therefore, these measurements recorded spatial and temporal variations of water temperature upstream of the dam. Water temperature was collected using a 95A semiconductor thermometer within ±0.1 °C of the true water temperature. The probe was cleaned and calibrated against a reference thermometer three or four times a year.
Figure 2

Spatial distribution of water temperature at the monitoring section (a) and temporal variation at the hydropower intake (b) for Danjiangkou reservoir.

Figure 2

Spatial distribution of water temperature at the monitoring section (a) and temporal variation at the hydropower intake (b) for Danjiangkou reservoir.

There were two hydrological stations in the upstream of Danjiangkou reservoir, i.e., Baihe and Zijingguan, and four hydrological stations in the study reach, i.e., Huangjiagang, Xiangyang, Zhuandouwan, and Nianpanshan, that recorded water temperature, water level, and flow at 8:00 a.m. each day during the period from 1969 to 1980 (Figure 3). Water temperature measurements were collected at the depth of 0.5 m near the base water gauge using a 591 frame type thermometer within ±0.2 °C of the true values. Water level was measured utilizing water gauge near the river bank and flow was measured utilizing current meter method. The reported monthly averages were calculated based on these records.
Figure 3

Monthly water temperature (a), flow (b), and water level (c) during the period from 1969 to 1980.

Figure 3

Monthly water temperature (a), flow (b), and water level (c) during the period from 1969 to 1980.

Modeling approach

Thermal structure in the Danjiangkou reservoir is modeled with a 3-D EFDC model, and stream temperatures in the downstream river are calculated using ETM model. Results (the release temperatures) from the EFDC model are used as the input of ETM model to estimate stream temperatures.

Reservoir water temperature model

The model is a general purpose modeling package for simulating 1-D, 2-D, and 3-D flow, transport, and biogeochemical processes, initially developed by Virginia Institute of Marine Science, and subsequently developed by the US Army Corps of Engineers sponsored by the US EPA (Jeong et al. 2010). It has been used extensively to simulate hydrodynamic and water quality variations in rivers, lakes, reservoirs, wetlands, estuaries, and coastal regions integrated with a water-column eutrophication model (Park et al. 2005; Luo & Li 2009). The EFDC model is selected in this study because it can well simulate thermal structure in deep reservoirs (Yang et al. 2012). Water temperature is modeled using the 3-D form of the advection–dispersion equation: 
formula
1
where T is water temperature (°C); u, v, and ω are velocity components (m s−1) in the curvilinear, orthogonal coordinates, x-, y-, and z-directions, respectively; Ax, Ay, and Az, are turbulent diffusivities (m2 s−1) in the x-, y-, and z-directions, respectively; ST is source and sink terms per unit volume (°C s−1); H is water column depth (m); and mx, my are horizontal curvilinear coordinate scale factors.
In Equation (1), on the left-hand side, the first term is local acceleration term and the last three terms are the advective transport terms; on the right-hand side, the first three terms account for the diffusive transport and the last term represents the kinetic processes and external loads for each of the state variables. The kinetic equation is solved using a second-order accurate trapezoidal Crank–Nicholson scheme (Park et al. 2005). The calculation of the heat fluxes (source terms) is based on the NOAA Geophysical Fluid Dynamic Laboratory's atmospheric heat exchange formulation (Rosati & Miyakoda 1988; Yang et al. 2012): 
formula
2
 
formula
3
 
formula
4
 
formula
5
 
formula
6
where Φ is the net surface heat flux; ΦI is the solar shortwave radiation; ΦB is the net longwave radiation; Φc is the convective heat flux; L is the latent heat of evaporation; Φe is the evaporation heat flux; I is the downward flux of solar radiation at the water surface (W m−2); F is a distribution fraction between 0 and 1, for thermal processes, which is determined by model calibration in the EFDC model; SF and SS are fast and slow scale attenuation coefficients (m−1), respectively; at the bottom, z = 0, and free surface, z = 1; ɛ is the emissivity (0.97); σ is Stefan–Boltzman constant (5.67 × 10−8 J m−2 s−1 K−4); Ts and Ta are the water surface and air temperatures (°C), respectively; ea is the vapor pressure (mb); Bc is an empirical constant equal to 0.8; Cc is the fractional cloud cover; ρa is the air density (1.2 × 10−3 g cm−3); ce and ch are dimensionless evaporation and convective transfer coefficient on the order of 10−3 in magnitude, respectively; Uw and Vw are the components of the wind velocity at 10.0 m above the water surface; ess and esa are the saturation vapor pressures (mb) corresponding to the water surface and air temperatures, respectively; Rh is the fractional relative humidity; pa is the surface air pressure (mb); and cp is the specific heat capacity (1.005 × 103 J kg−1 K−1).

Stream temperature model

ETM is a physically based model, which estimates stream temperatures using equilibrium temperature theory on the basis of air temperature, flow volume, water temperature, and stream morphology information. The model derives from the advection–dispersion-heat balance equations after assuming steady state condition, complete vertical and lateral mixing in river cross sections, advection dominating over dispersion (longitudinal) (Walters et al. 2000; Wright et al. 2009) as follows: 
formula
7
where T is water temperature (°C); Q is river discharge (m3 s−1); A is cross-sectional area (m2); D is mean flow depth (m); ρ is water density (1,000 kg m−3); Cp is heat capacity of water (4,185 J kg−1 °C−1); and Hf is the net rate of heat exchange across the water surface (W m−2). This net heat flux can be linearized as a function of equilibrium temperature by assuming that the net rate of heat exchange is proportional to the departure from the temperature equilibrium (Bustillo et al. 2014): 
formula
8
where Te is equilibrium temperature (°C), defined as the water temperature at which the net rate of heat exchange at the water body interface is 0. Ke is an overall heat exchange coefficient (W m−2 °C−1) related to air temperature, wind speed, and relative humidity. Substituting Hf from Equation (8) into Equation (7) and solving the differential equation gives an equation for stream temperature: 
formula
9
where T0 is the temperature at the upstream boundary (°C).
Many previous studies to model river water temperature using the equilibrium temperature concept have considered Ke as a constant, yielding satisfactory model performance (Marće & Armengol 2008; Bustillo et al. 2014). Therefore, we also considered Ke as a single-valued adjustable parameter in this study. By contrast, Te cannot be considered as a constant, so we calculated Te based on the linear relationship between equilibrium and air temperature (Ta) (Caissie et al. 2005): 
formula
10
where ae and be are regression coefficient parameters between Te and Ta. Figure 4 shows a comparison between monthly water temperature and monthly air temperature at Huangjiagang and Nianpanshan stations. The correlation is stronger at the Nianpanshan station, which indicates that monthly water temperature is quite close to monthly equilibrium temperature at the monitoring position far away from the dam. Therefore, we can determine the initial values of ae and be through the linear relationship between downstream water temperatures, which are not affected by the dam, with air temperature. The final parameters can be determined from the observed data.
Figure 4

Comparison of monthly water temperature and monthly air temperature at Huangjiagang station (a) and Nianpanshan station (b).

Figure 4

Comparison of monthly water temperature and monthly air temperature at Huangjiagang station (a) and Nianpanshan station (b).

Numerical grid

In the study, the EFDC model uses the horizontally curvilinear orthogonal grid. Grid sensitivity analysis indicates that model performance is better with finer grid. However, this would inevitably result in more computational cost. As a trade-off between the accuracy and efficiency, the grid map in Figure 5 is selected. It consists of 1,500 cells horizontally, with the cell size ranging from 0.2 to 1.7 km. Furthermore, 40 layers are considered vertically.
Figure 5

Grid map for Danjiangkou reservoir.

Figure 5

Grid map for Danjiangkou reservoir.

Model initialization

The EFDC model operates on a 4-minute time step, although daily input and output data are used. Initial conditions of the model include surface water level at the reservoir, and boundary conditions include reservoir inflow and water temperature at inlet, and outflow at outlet. The position of inlet boundary is determined by reservoir flood analysis. A uniform velocity distribution is assumed at the inlet. Dirichlet boundary condition is used for flow field calculation at the inlet and outlet boundary, and water temperature calculation at the inlet boundary. In detail, the discharge and water temperature at Baihe and Zijingguan hydrological stations are used as the inlet boundary condition of the upper Han River and Dan River, respectively. As well, the release discharge of Danjiangkou reservoir is used as the outlet boundary condition to provide the information of different outflow positions in the vertical direction, including hydropower intake and overflow weir. The calculated output temperatures include the temperature of hydropower intake and overflow weir. These temperatures are obtained through taking an average of corresponding vertical layer at two locations.

The ETM model is implemented on a monthly time scale along the river which is discretized by 1 km cells in this study. Boundary conditions of the ETM model include stream inflow and water temperature at inlet. In detail, the calculated release temperature and observed release discharge are used as the inlet boundary condition. The release temperature of the Danjiangkou reservoir is on a daily time scale and hence its average monthly values are finally used as input of the ETM model.

Simulation scenarios and input data

To quantitatively analyze the influence of the large water diversion projects, the Danjiangkou dam heightening project, and the two small reservoirs downstream, on the thermal structures in the Danjiangkou reservoir and its downstream river, four scenario simulations are carried out using a multi-model approach, including EFDC and ETM models. The first scenario (S1) is the reference situation, which does not consider the effect of any of the projects. The second scenario (S2) just considers the effect of heightening the Danjiangkou dam. Both dam heightening project and large water diversion projects are taken into account in the third scenario (S3). The fourth scenario (S4) is comprehensive, considering the influence of all the above projects, including the project to heighten the Danjiangkou dam, large water diversion projects, and two small reservoirs downstream.

The inputs of the two models include meteorological data, hydrological data, and water temperature. For different scenarios, the meteorological data are the same, while the other two are different. The meteorological input and the simulated output of the two models are shown in Table 2, while the hydrological and water temperature inputs at the four scenarios are shown in Table 3. For the EFDC model, all these data are measured and needed as inputs. As well, these hydrological and water temperature data are regarded as the initial and boundary conditions of the EFDC model. The values of the boundary conditions used at four simulation scenarios are shown in Figure 6. For the ETM model, the inflow water temperature is the reservoir release temperature calculated by EFDC model, and stream inflow discharge and the normal water levels at two small reservoirs are measured. Table 3 and Figure 6 show that the main differences are the surface water elevation in the Danjiangkou reservoir between S1 and S2, the inflow and outflow of the reservoir between S2 and S3, and the normal water level of Wangfuzhou and Cuijiaying reservoirs between S3 and S4. Noticeably, the outflow of EFDC model at S3 and S4 includes the release water from the Danjiangkou reservoir and diversion water from the South-to-North Water Transfer Project and Ebei Water Transfer Project. For S4, as the two small reservoirs have typical river-type, low-head stream-flow hydroelectric plants, it is assumed that the area is vertically mixing and thus simulated by a 1-D analytical solution.
Table 2

Model meteorological input and simulated output

Model Meteorological input Simulated output 
EFDC Daily air pressure, air temperature, percent relative humidity, precipitation, wind speed, wind direction, evaporation, total incident solar radiation, cloud cover Reservoir water temperature 
ETM Monthly air temperature Stream water temperature 
Model Meteorological input Simulated output 
EFDC Daily air pressure, air temperature, percent relative humidity, precipitation, wind speed, wind direction, evaporation, total incident solar radiation, cloud cover Reservoir water temperature 
ETM Monthly air temperature Stream water temperature 
Table 3

Hydrological and water temperature inputs for the EFDC and ETM models at the four simulation scenarios

Scenario EFDC ETM 
S1 Inflowaqin1 Inflow qin1 = qout1 
Inflow water temperaturebTin
Outflow qout1 
Inflow water temperature Tin1 = Tout1 
Surface water level hs1  
S2 Inflow qin1 Inflow qin2 = qout2 
Inflow water temperature Tin
Outflow qout2 
Inflow water temperature Tin2 = Tout2 
Surface water level hs2  
S3 Inflow qin2 = qin1qd1c Inflow qin3 = qout3 
Inflow water temperature Tin Inflow water temperature Tin3 = Tout3 
Outflow qout3 = qout2qd2dqd3e  
Outflow qd2d  
Outflow qd3e  
Surface water level hs2  
S4 Same as S3 Inflow qin3 
 Inflow water temperature Tin3 
 Normal water level hn1 in Cuijiaying reservoir and hn2 in Wangfuzhou reservoir 
Scenario EFDC ETM 
S1 Inflowaqin1 Inflow qin1 = qout1 
Inflow water temperaturebTin
Outflow qout1 
Inflow water temperature Tin1 = Tout1 
Surface water level hs1  
S2 Inflow qin1 Inflow qin2 = qout2 
Inflow water temperature Tin
Outflow qout2 
Inflow water temperature Tin2 = Tout2 
Surface water level hs2  
S3 Inflow qin2 = qin1qd1c Inflow qin3 = qout3 
Inflow water temperature Tin Inflow water temperature Tin3 = Tout3 
Outflow qout3 = qout2qd2dqd3e  
Outflow qd2d  
Outflow qd3e  
Surface water level hs2  
S4 Same as S3 Inflow qin3 
 Inflow water temperature Tin3 
 Normal water level hn1 in Cuijiaying reservoir and hn2 in Wangfuzhou reservoir 

aIncluding inflow from Baihe and Zijingguan stations.

bIncluding inflow water temperature from Baihe and Zijingguan stations.

cDiversion water from Han River-to-Wei River Water Transfer Project qd1.

dDiversion water from South-to-North Water Transfer Project qd2.

eDiversion water from Ebei Water Transfer Project qd3.

Figure 6

Boundary conditions for the EFDC model: (a) inflow for S1 and S2 only, with a small difference to S3 and S4; (b) inflow temperature for four simulation scenarios; and (c) outflow for four simulation scenarios.

Figure 6

Boundary conditions for the EFDC model: (a) inflow for S1 and S2 only, with a small difference to S3 and S4; (b) inflow temperature for four simulation scenarios; and (c) outflow for four simulation scenarios.

Model calibration and validation

The EFDC model parameters, mainly for thermal processes were calibrated by field data of water temperature collected from the Danjiangkou reservoir during three hydrological years, a wet year (04/1974–03/1975), a normal year (04/1970–03/1971), and a dry year (04/1977–03/1978). Water temperature measurements were compared with the simulated values at the corresponding vertical layer. Hydrological guarantee rate of the flow for the three hydrological years were P = 25%, P = 50%, and P = 75%, respectively. A comparison of the meteorological conditions is shown in Figure 7. The intra-annual variation patterns of the precipitation and air temperature were similar, with heavier rain and higher temperature in summer than in winter. The mean annual temperature was 15.0 °C, 15.1 °C, and 15.5 °C for wet, normal, and dry year, respectively, indicating small temperature differences among the three hydrological years. However, annual precipitation showed very significant differences, and the values were 1,014, 764, and 526 mm in the 3 years. This suggested that the selected hydrological years could represent both average and extreme weather situations. The model was validated with the measured temperature data for another three hydrological years, a wet year (04/1975–03/1976), normal year (04/1973–03/1974), and dry year (04/1978–03/1979).
Figure 7

Meteorological conditions: precipitation (a) and air temperature (b) for three hydrological years.

Figure 7

Meteorological conditions: precipitation (a) and air temperature (b) for three hydrological years.

Equation (9) used river morphology information to calculate the downstream water depth (D) and water temperature (T). Cross sections at Huangjiagang and Nianpanshan hydrological stations and river bed elevation of 1 km resolution along the study reach were available regarding river morphology. This led to the empirical relationships used thereafter that relate the mean river depth (D) with the river discharge (Q): D =3.03 Q0.11 (R2 = 0.95) and mean flow velocity (v) with the river discharge (Q): v =0.0013 Q0.81 (R2 = 0.98). These relationships could then be used to calculate river water temperature. The ETM model used the same calibration and validation periods as the EFDC model. During the periods, hydrological data were used as inputs for the stream temperature model. Monthly water temperatures at four hydrological stations (Huangjiagang, Xiangyang, Zhuandouwan, and Nianpanshan) were used to calibrate the model parameters, regression coefficient parameters ae and be, and heat exchange coefficient Ke, and validate the modeling results.

Model performance was assessed with mean signed error (MSE), mean absolute error (MAE), and root mean square error (RMSE): 
formula
11
 
formula
12
 
formula
13
where Pi and Qi are predicted and observed values of water temperature at the Danjiangkou reservoir and its downstream river, and n is the number of observations.

Scale factor and recuperation distance

Equation (9) can be expressed as: 
formula
14
with 
formula
15
where Cx is the scale factor which reflects the influence of T0 and Te on stream water temperature.

Equation (14) can quantitatively distinguish the impact of the release temperature of the dam (CxT0) and the equilibrium temperature (1—Cx)Te. The former part reflects the effect of human disturbance from the dam, while the latter represents the influence of meteorological conditions, as equilibrium temperature is a function of air temperature. This study mainly focuses on the effect of human disturbance (CxT0), and therefore the meteorological conditions remain constant for different scenarios.

For a smaller Cx, the influence of Te is much larger according to Equation (14), which indicates that it is much easier to reach natural state at a given position (x). Therefore, Cx can be regarded as an indicator of recovery rate at this position, and for a given river reach, ∑Cx/x can be regarded as an indicator of its recovery rate. From Equation (15), it is seen that Cx is related to distance (x), depth (D), area (A), and discharge (Q). For a given river morphology, D, A, and Q are related and can be converted to each other. Therefore, A/DQ can be assumed as 1/Dv, where v is velocity. This indicates that a change of river discharge will affect Cx at a given position and then influence the recovery rate of the stream temperature.

As Cx can reflect the influence of river discharge on recovery rate, it can also be used to calculate the recuperation distance. We assume that the recuperation distance is the distance that reduces the influence of release temperature to less than or equal to 5% (i.e., Cx ≤5%). The differences between T0 and Te are less than 12 °C across the different simulation scenarios, and then the temperature variations outside the recuperation distance are less than 0.6 °C, which reflects a similar effect to Prats et al.’s (2012) definition.

Ecological impact assessment method

In this study, fish are used as a marker of the ecological impact due to water temperature variations. Two indexes, the time to reach the minimum threshold for spawning and spawning opportunities index are selected to assess this influence. The former can be obtained through comparing the minimum water temperature requirement for spawning with the water temperature in the river during the spawning period. The latter, which is used to quantify the opportunities of the fish to successfully spawn, can be calculated with the following equation according to Preece & Jones (2002): 
formula
16
where Tw is the duration of the whole spawning period and Ta is the duration of stream temperature above the minimum water temperature requirement for spawning.

RESULTS

Model performance

Figure 8 compares model results against temperature measurements and Table 4 shows the number of observations (n) of the Danjiangkou reservoir and the error statistics during the calibration and validation periods. Figure 8 shows that the model provides very good estimations for different water depth in each period. Overall, the MSEs suggest a good agreement between the model and observations with maximum error in the 0.25 °C range. MAEs are less than 0.72 °C and RMSEs are within 0.52 °C under various hydrological situations. This indicates that the EFDC model can successfully capture the timing and depth of thermal stratification.
Table 4

Calibration and validation error statistics of daily water temperature by Danjiangkou reservoir at different hydrological years

 Wet year Normal year Dry year 
Calibration error statistics 
n 153 146 143 
 MSE, °C 0.15 − 0.13 − 0.11 
 MAE, °C 0.48 0.50 0.48 
 RMSE, °C 0.42 0.46 0.39 
 Calibration period (month/year) 04/1974–03/1975 04/1970–03/1971 04/1977–03/1978 
Validation error statistics 
n 137 165 123 
 MSE, °C 0.16 − 0.25 0.16 
 MAE, °C 0.53 0.72 0.54 
 RMSE, °C 0.45 0.52 0.41 
 Validation period (month/year) 04/1975–03/1976 04/1973–03/1974 04/1978–03/1979 
 Wet year Normal year Dry year 
Calibration error statistics 
n 153 146 143 
 MSE, °C 0.15 − 0.13 − 0.11 
 MAE, °C 0.48 0.50 0.48 
 RMSE, °C 0.42 0.46 0.39 
 Calibration period (month/year) 04/1974–03/1975 04/1970–03/1971 04/1977–03/1978 
Validation error statistics 
n 137 165 123 
 MSE, °C 0.16 − 0.25 0.16 
 MAE, °C 0.53 0.72 0.54 
 RMSE, °C 0.45 0.52 0.41 
 Validation period (month/year) 04/1975–03/1976 04/1973–03/1974 04/1978–03/1979 
Figure 8

Comparison of observed and simulated daily water temperature during the calibration period, including a wet year (a), a normal year (b), and a dry year (c); and validation period, including another wet year (d), normal year (e), and dry year (f) at Danjiangkou reservoir.

Figure 8

Comparison of observed and simulated daily water temperature during the calibration period, including a wet year (a), a normal year (b), and a dry year (c); and validation period, including another wet year (d), normal year (e), and dry year (f) at Danjiangkou reservoir.

Figure 9 shows the comparison between the observed and predicted temperatures, and Table 5 contains the error statistics for the sites in the study reach within the calibration and validation data set. The ETM model provides very good estimations and follows the water temperature variations closely (Figure 9). The MSEs are close to zero (−0.13–0.09 °C), indicating no consistent positive or negative bias. For all sites, the MAEs and RMSEs are less than 0.52 °C regardless of year. This suggests that the model is capable of simulating water temperature in the study river reaches.
Table 5

Calibration and validation error statistics of monthly water temperatures for each monitoring location in study reach at different hydrological years

    Wet year
 
Normal year
 
Dry year
 
Site Distance, km MSE, °C MAE, °C RMSE, °C MSE, °C MAE, °C RMSE, °C MSE, °C MAE, °C RMSE, °C 
Calibration error statistics 
 Huangjiagang 0.01 0.19 0.23 − 0.04 0.22 0.28 − 0.04 0.22 0.16 
 Xiangyang 111 0.04 0.38 0.29 0.01 0.36 0.28 0.03 0.38 0.24 
 Zhuandouwan 209 − 0.06 0.43 0.29 − 0.05 0.48 0.36 − 0.03 0.47 0.27 
 Nianpanshan 249 0.06 0.45 0.33 − 0.04 0.49 0.39 − 0.02 0.49 0.29 
 Calibration period (month/year) 04/1974–03/1975 04/1970–03/1971 04/1977–03/1978 
Validation error statistics 
 Huangjiagang 0.01 0.26 0.24 0.05 0.25 0.18 − 0.06 0.34 0.28 
 Xiangyang 111 0.01 0.44 0.36 0.06 0.40 0.26 0.09 0.44 0.25 
 Zhuandouwan 209 − 0.04 0.49 0.39 0.01 0.52 0.41 − 0.08 0.51 0.27 
 Nianpanshan 249 − 0.09 0.50 0.40 − 0.07 0.52 0.51 − 0.13 0.51 0.37 
 Validation period (month/year) 04/1975–03/1976 04/1973–03/1974 04/1978–03/1979 
    Wet year
 
Normal year
 
Dry year
 
Site Distance, km MSE, °C MAE, °C RMSE, °C MSE, °C MAE, °C RMSE, °C MSE, °C MAE, °C RMSE, °C 
Calibration error statistics 
 Huangjiagang 0.01 0.19 0.23 − 0.04 0.22 0.28 − 0.04 0.22 0.16 
 Xiangyang 111 0.04 0.38 0.29 0.01 0.36 0.28 0.03 0.38 0.24 
 Zhuandouwan 209 − 0.06 0.43 0.29 − 0.05 0.48 0.36 − 0.03 0.47 0.27 
 Nianpanshan 249 0.06 0.45 0.33 − 0.04 0.49 0.39 − 0.02 0.49 0.29 
 Calibration period (month/year) 04/1974–03/1975 04/1970–03/1971 04/1977–03/1978 
Validation error statistics 
 Huangjiagang 0.01 0.26 0.24 0.05 0.25 0.18 − 0.06 0.34 0.28 
 Xiangyang 111 0.01 0.44 0.36 0.06 0.40 0.26 0.09 0.44 0.25 
 Zhuandouwan 209 − 0.04 0.49 0.39 0.01 0.52 0.41 − 0.08 0.51 0.27 
 Nianpanshan 249 − 0.09 0.50 0.40 − 0.07 0.52 0.51 − 0.13 0.51 0.37 
 Validation period (month/year) 04/1975–03/1976 04/1973–03/1974 04/1978–03/1979 
Figure 9

Comparison of observed and simulated monthly water temperature during calibration period (a) and validation period (b).

Figure 9

Comparison of observed and simulated monthly water temperature during calibration period (a) and validation period (b).

The performance of the EFDC model in validation periods is slightly less satisfactory relative to calibration periods due to the inaccuracy of parameters. For the ETM model, MAEs and RMSEs tend to increase downstream during both calibration and validation periods, because the effect of the tributaries in the model is not considered.

Thermal structure variation in the Danjiangkou reservoir

Elevation–time isotherms (Figure 10) indicate the Danjiangkou reservoir's seasonal thermal cycle. Thermal structures in the reservoir were similar across both hydrological years and simulation scenarios. Thermal structure of the Danjiangkou reservoir at S1 in the normal year is a representative example (Figure 10(b)). The reservoir was approximately isothermal from November to March, before seasonal stratification began in April with heating of the surface layer. A distinct thermocline formed between May and October with a thermal gradient from 0.3 to 0.36 °C. The temperature of the surface layer rose from 8.0 °C in February to 28 °C in August. An increase in temperature was also evident in the hypolimnion, with the average monthly temperature rising from an annual minimum of 8.0 °C in February to 16 °C in September. The most obvious difference in thermal structures among the three hydrological conditions is that the thermocline minimum (isotherm 18 °C) in the dry year was much lower than in the wet year, about 7 m below (Figure 10(a) and 10(c)), which means that surface meteorological conditions can influence the water body close to the bottom of the reservoir when the water surface is low enough in a dry year.
Figure 10

Elevation–time isotherm diagrams for Danjiangkou reservoir: (a) wet year, S1; (b) normal year, S1; (c) dry year, S1; (d) normal year, S2; (e) normal year, S3 and S4.

Figure 10

Elevation–time isotherm diagrams for Danjiangkou reservoir: (a) wet year, S1; (b) normal year, S1; (c) dry year, S1; (d) normal year, S2; (e) normal year, S3 and S4.

At S2, the thermal gradient of the thermocline decreases to 0.21–0.32 °C m−1 and the position of the 18 °C isotherm rises by about 2 m relative to S1 in normal year, which may be caused by the increased water depth weakening the heat exchange with the upper water body. In addition, the intra-annual amplitude of bottom temperature decreases by 2 °C. This may be also related to the deeper water, which will increase the thermal capacity and decrease the thermal inertia. The thermal structures at S3 and S4 are identical due to the same boundary conditions in Danjiangkou reservoir, shown in Figure 10(e). The positions of the 20, 22, and 24 °C isotherms in the thermocline are higher at S3 and S4 than S2, which will affect the release temperature, as the first two isotherms are located within the hydropower intake.

Reservoir release temperature variation

Figure 11(a) shows the release temperature variation, including hydropower intake temperature and surface temperature, at S1 across three hydrological years. Temperature variations at surface layer are similar in the different hydrological years, and all follow the change of air temperature, which rises from April, reaches a maximum in August, and then falls. This indicates that the surface temperature is mainly affected by the meteorological conditions. Intra-annual temperature variation in hydropower intake lags behind the surface temperature, and the maximum temperature appears in September, except in the dry year. The temperatures in hydropower intake are lower than the surface for thermal stratification between May and September. Across the hydrological situations, intra-annual amplitudes are greater in the dry year, about 16.6 °C, than in the wet year, about 15.4 °C.
Figure 11

Release temperature from the Danjiangkou reservoir according to S1 (a), S2 (b), and S3 and S4 (c) in three hydrological years.

Figure 11

Release temperature from the Danjiangkou reservoir according to S1 (a), S2 (b), and S3 and S4 (c) in three hydrological years.

Figure 11(b) demonstrates the release temperature variation at S2 across the hydrological years. The hydropower intake temperatures are around 1.3–2.4 °C lower in summer and around 0.7–2.8 °C higher in winter, relative to S1. Their intra-annual temperature variation lags behind, for example, the maximum appears in September in the dry year. The difference of hydropower intake temperatures between S2 and S3 is very small in winter, around 0.1–0.2 °C, but becomes obvious in summer, around 0.6–0.8 °C (Figure 11(b) and 11(c)).

Overall, the difference in the release temperature is not significant for different hydrological conditions, which demonstrates that the effect of the hydrological conditions on the release temperature is limited. Therefore, the following analysis is mainly based on the normal year.

Stream temperature variation

Figure 12(a) shows the stream temperature change at S1. The warm period (above 28 °C) is mainly in the summer (June to August), and the cold period (below 6 °C) is mainly in the winter (December to February), which is similar to the variation of air temperature. Downstream from the Danjiangkou dam, water temperature tends to gradually recover to its natural state. The temperature changes at S2 are presented in Figure 12(b). The release temperatures of the Danjiangkou dam are lower in summer and higher in winter relative to S1. The high temperature period is confined to between July and August, which is shorter than S1 (June to August). The distance of the reaches with high temperature is approximately 400 km, which is also less than S1 (about 470 km). Figure 12(c) demonstrates temperature variations at S3. The distances of river channel with high and low temperatures are much longer than S2, by 60 km and 80 km, respectively. This is because the release discharge is decreased by the water diversion projects, about 345 m3 s−1, which will reduce thermal inertia and increase the recovery rate of water temperature. As well, the ‘temperature valley’ in July (about 100–350 km) is related to a high volume and slow recovery rate. At S4, the temperature increases about 0.5 °C in summer and decreases about 0.3 °C in winter in Cuijiaying reservoir from 102 to 134 km, compared to S3 (Figure 12(d)). This is because the river has a slower velocity there and longer exposure to atmospheric conditions. The study suggests that these projects will alter the thermal regime of the river in both space and time.
Figure 12

Monthly stream temperatures in the study reach at S1 (a), S2 (b), S3 (c), and S4 (d) during three hydrological years.

Figure 12

Monthly stream temperatures in the study reach at S1 (a), S2 (b), S3 (c), and S4 (d) during three hydrological years.

Figure 13 represents the longitudinal stream temperature profiles for the four simulation scenarios in August. At S2, the release temperature decreases by 4.0 °C relative to S1, which leads to the reduction of the stream temperature near the dam and an improvement in the recovery rate from 0.015 to 0.033 °C km−1 (from the Danjiangkou dam to Cuijiaying dam). Further at S3, the release temperature decreases by 1.4 °C compared with S2, and the recovery rate increases to 0.049 °C km−1. This increase in the recovery rate results from the reduced flows – 1,240 at S1, 954 at S2, and 608 m3 s−1 at S3 – indicating that less water needs to be heated. Finally, considering the Wangfuzhou and Cuijiaying reservoirs (S4), recovery rates increase in their backwater areas (0–30 km and 102–134 km). For example, recovery rate increases about 0.005 °C km−1 at the region from 102 to 134 km compared with S3.
Figure 13

Longitudinal stream temperature profiles at the four simulation scenarios in August.

Figure 13

Longitudinal stream temperature profiles at the four simulation scenarios in August.

DISCUSSION

This section first discusses how the release discharge and release temperature in the Danjiangkou reservoir affects the thermal regime in the downstream river. Herein, the effect of the release discharge is analyzed through a scale factor, which can reflect the change of the recovery rate and then can be applied to calculate the recuperation distance for the dam. Thereafter, the impact of the large water diversion projects on the release discharge and release temperature is investigated. Finally, ecological impact of water temperature change is studied based on fish spawning.

Scale factor and the release discharge

Figure 14 compares the longitudinal scale factor (Cx) variations for S1, S2, and S3 in August. The scale factors are 0.78, 0.75, and 0.65 for S1, S2 and S3, respectively, for the reach from the Danjiangkou dam to the Cuijiaying dam. This suggests that the effect of human disturbance shown in S2 and S3 is much weaker than that in S1. Previous analysis shows that the corresponding recovery rates are 0.015 °C km−1 for S1, 0.033 °C km−1 for S2, and 0.049 °C km−1 for S3, which can further confirm the analysis. In addition, the recuperation distances of the first three scenarios are 592, 542, and 359 km, respectively. This all results from higher flows – 1,240 in S1, 954 in S2, and 608 m3 s−1 in S3, indicating that a reduction in river discharge will decrease Cx value, weaken the human influence, improve recovery rate, and reduce the recuperation distance. In turn, it also indicates that the recuperation distance in the study can be used to quantitatively compare water temperature variations due to the change of the release discharge. From the aspect of the whole year, the recuperation distances of the three scenarios are 616 km, 758 km, and 578 km, respectively, in summer and 597 km, 500 km, and 314 km, respectively, in winter, showing that the recuperation distance reduces 38 km in summer and 283 km in winter due to the Danjiangkou dam heightening project and large water diversion projects.
Figure 14

Comparison of longitudinal scale factor (Cx) profiles at S1, S2, and S3.

Figure 14

Comparison of longitudinal scale factor (Cx) profiles at S1, S2, and S3.

The release temperature

In addition to the release discharge, the water temperature in the downstream river is influenced by the release temperature. The temperature change caused by the release temperature is CxT0 (Equation (14)), where △T0 is the variation of the release temperature. Figure 15 shows the longitudinal stream temperature profile pre- and post-dam heightening in August with a constant release discharge and variable release temperature. In the Wangfuzhou and Cuijiaying dams, the temperatures decrease by about 3.5 and 2.2 °C with a reduction of the release temperature by 4 °C from the Danjiangkou dam. As Cx reduces longitudinally, the influence of the release temperature also weakens along the river, which indicates that the release temperature mainly affects the area near the dam. As well, the recovery rate of the river reach from Danjiangkou dam to Cuijiaying dam rises from 0.015 to 0.029 °C km−1 due to decreasing release temperature.
Figure 15

Comparison of longitudinal stream temperature profiles in August pre- and post-dam heightening with the same constant release discharge and varying release temperature (solid and dotted lines represent pre- and post-dam heightening conditions, respectively).

Figure 15

Comparison of longitudinal stream temperature profiles in August pre- and post-dam heightening with the same constant release discharge and varying release temperature (solid and dotted lines represent pre- and post-dam heightening conditions, respectively).

The large water diversion projects

In the study, the impacts of large water diversion projects are considered from two aspects. On the one hand, the projects reduce release discharge, weaken thermal intra, improve recovery rate, and then increase water temperature variability. For example, the intra-annual temperature range of the river reach from 180 to 250 km widens from around 6.3–25.5 to 5.2–26.5 °C. On the other hand, they change thermal structure, reduce release temperature in the reservoir, and decrease water temperature in the downstream river in summer. Longitudinal stream temperature profile at S2′ in Figure 16 is calculated based on the release temperature at S3 and release discharge at S2 in August. The difference between S2 and S2′ can reflect the influence of the release temperature change. At S2′, release temperature decreases by 1.4 °C due to the large water diversion projects, which reduces the temperature at least 0.7 °C near the dam (Cx ≤ 50%) relative to S2. However, the maximum difference of temperature between S3 and S2 in this area is about 1.4 °C, which suggests that the effect due to the change of release temperature cannot be ignored when analyzing the influence of water diversion projects.
Figure 16

Comparison of longitudinal stream temperature profiles at S2, S2′, and S3 in August.

Figure 16

Comparison of longitudinal stream temperature profiles at S2, S2′, and S3 in August.

Ecological impact

Herein, the spawning of four major Chinese carp species is used as a marker of ecological impact. A qualitative assessment of spawning opportunities was made by comparison of the water temperature against the minimum threshold during the spawning period (Preece & Jones 2002; Figure 17). The spawning opportunities index and the time to reach the threshold at S1 versus S4 at different locations are shown in Table 6. For all the reaches in the mid-Han River, the time to reach the minimum threshold at S4 shows a lag relative to S1, e.g., about 1.87 months at the Huangjiagang station. Furthermore, the spawning opportunities of fish in S4 are higher when the distance from the Danjiangkou dam is greater. This may result in a reduction of spawning sites, and the optimal spawning locations moving downstream from Huangjiagang (6 km) to Xiangyang (111 km).
Table 6

Spawning opportunity index and the time to reach the minimum threshold at S1 and S4 for four major Chinese carp species in the mid-Han River at different positions

Distance, km Site Spawning opportunity index, %
 
Time, month
 
S1 S4 S1 S4 
Huangjiagang 100 36 5.50 7.37 
10  100 38 5.48 7.32 
20  100 45 5.44 7.13 
40  100 80 5.35 6.19 
70  100 99 5.12 5.68 
111 Xiangyang 100 100 4.94 5.17 
249 Zhongxiang 100 100 4.52 4.57 
Distance, km Site Spawning opportunity index, %
 
Time, month
 
S1 S4 S1 S4 
Huangjiagang 100 36 5.50 7.37 
10  100 38 5.48 7.32 
20  100 45 5.44 7.13 
40  100 80 5.35 6.19 
70  100 99 5.12 5.68 
111 Xiangyang 100 100 4.94 5.17 
249 Zhongxiang 100 100 4.52 4.57 
Figure 17

Fish spawning period and minimum spawning temperature in relation to water temperature in S1 (a) and S4 (b) for different positions in the mid-Han River.

Figure 17

Fish spawning period and minimum spawning temperature in relation to water temperature in S1 (a) and S4 (b) for different positions in the mid-Han River.

CONCLUSIONS

Overall, the large water diversion projects and related auxiliary projects, including a project to heighten the Danjiangkou dam and two small reservoirs downstream in central China will change the spatio-temporal distribution of water temperature in the Danjiangkou reservoir and its downstream river. The Danjiangkou dam heightening project alters water temperature seasonally, increasing the temperature at hydropower intake around 0.7–2.8 °C in winter and decreasing it around 1.3–2.4 °C in summer. However, the water division projects lead to an improved temperature recovery rate and a reduction in recuperation distance, the former in August rising from 0.015 to 0.049 °C km−1 (from Danjiangkou dam to Cuijiaying dam), and the latter falling by 38 km in summer and 283 km in winter. Furthermore, two small reservoirs make water temperature recover much faster and further weaken the influence of the Danjiangkou dam heightening project, increasing the temperature about 0.5 °C in summer and decreasing it about 0.3 °C in winter in the backwater region of the Cuijiaying reservoir. Analysis of the respective effects due to each project will guide project operation and water resources management in the near future.

Water temperature alterations in the downstream river due to these projects depend greatly on the release temperature and the release discharge from the Danjiangkou reservoir. Change in the release temperature with the dam heightening project mainly affects the area near the dam. A reduction in river discharge with water diversion projects will decrease Cx value, weaken the effect of the release temperature, improve recovery rate, and reduce the recuperation distance. Based on the traditional studies that focus only on the discharge variation in water diversion, this study also considers the change of the thermal structure and the release temperature in the reservoir. As the influence of the release temperature is very obvious near the dam, it cannot be ignored in analysis of the effect of water diversion projects on the thermal regime of the downstream river.

The projects may have an adverse impact on the ecological environment as they alter the water temperature in the downstream river. The spawning periods for four major Chinese carp species may lag behind as the time to reach the minimum threshold shows a lag relative to the natural condition. Furthermore, the optimal spawning sites may move down along the river as the spawning opportunities of fish are higher when the distance from the Danjiangkou dam is greater. Therefore, some remedial work in the area should be carried out based on the finding of this study in order to weaken the adverse ecological impact.

ACKNOWLEDGEMENTS

The study is supported by the ‘948 Project’ of the Ministry of Water Resources of China (Grant No. 201434). The authors would like to thank Yangtze River Water Resources Commission and National Climate Centre of China Meteorological Administration for providing the data. The authors have declared no conflict of interest.

REFERENCES

REFERENCES
Albek
M.
Albek
E.
2009
Stream temperature trends in Turkey
.
Clean Soil Air Water
37
,
142
149
.
Bartholow
J. M.
Campbell
S. G.
Flug
M.
2004
Predicting the thermal effects of dam removal on the Klamath River
.
Environ. Manage.
34
,
856
874
.
Cluis
D. A.
1972
Relationship between stream water temperature and ambient air temperature
.
Nord. Hydrol.
3
,
65
71
.
Hamrick
J. M.
1992
A Three-Dimensional Environmental Fluid Dynamics Computer Code: Theoretical and Computational Aspects
.
Special Report in Applied Marine Science and Ocean Engineering; no. 317
.
Virginia Institute of Marine Science College of William and Mary
,
Gloucester Point, VA
.
Li
X. F.
Huang
D. M.
Xie
W. X.
Xie
S.
Chang
X.
Yang
H. Y.
Zhang
Y. Q
He
J. Q.
2006
Current status of spawning grounds of fish with pelagic eggs in the middle reaches of Hanjiang River
.
J. Dalian Fisheries University
21
,
105
111
(in Chinese).
Meier
W.
Bonjour
C.
Wüest
A.
Reichert
P.
2003
Modeling the effect of water diversion on the temperature of mountain streams
.
J. Environ. Eng.
129
,
755
764
.
Null
S. E.
Ligare
S. T.
Viers
J. H.
2013
A method to consider whether dams mitigate climate change effects on stream temperatures
.
J. Am. Water Resour. Ass.
49
,
1456
1472
.
Politano
M.
Haque
MD. M.
Weber
L. J.
2008
A numerical study of the temperature dynamics at McNary Dam
.
Ecol. Modell.
212
,
408
421
.
Rosati
A.
Miyakoda
K.
1988
A general circulation model for upper ocean simulation
.
J. Phys. Ocean
18
,
1601
1626
.
Shi
F. F.
Huang
W.
2009
Ecological impacts of the Danjiagnkou reservoir on the middle and lower reaches of the Han River
.
Resources and Environment in the Yangtze Basin
18
,
954
958
(in Chinese).
Smith
K.
1975
Water temperature variations within a major river system
.
Nord. Hydrol.
6
,
155
169
.
Walters
C.
Korman
J.
Stevens
L. E.
Gold
B.
2000
Ecosystem modeling for evaluation of adaptive management policies in the Grand Canyon
.
Conserv. Ecol.
4
,
1
.
Wright
S. A.
Anderson
C. R.
Voichick
N.
2009
A simplified water temperature model for the Colorado River below Glen Canyon Dam
.
River. Res. Appl.
25
,
675
686
.